Reproduce “Dimension-Free Decentralized Nonsmooth Nonconvex Stochastic Optimization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuanyu Wan, Lan Xue, Haomin Bai, Tong Wei, Mingli Song
The authors develop a decentralized online-to-nonconvex method for nonsmooth, nonconvex stochastic optimization across networked nodes. A shared random interpolation weight helps transfer consensus from decentralized online learning, while smoothing supports gradient comparisons. Under the stated assumptions, the method finds an (δ, ε)-Goldstein stationary point in expectation with O(δ⁻¹ε⁻³) stochastic-gradient queries and communication complexity with logarithmic, rather than polynomial, dimension dependence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.